Toward Understanding the Influence of Individual Clients in Federated Learning
This work addresses the problem of understanding individual client contributions in federated learning, which is important for debugging and improving model fairness for practitioners and researchers in distributed machine learning.
This paper introduces Fed-Influence, a new metric to quantify the influence of individual clients on the global model parameters in federated learning. They developed an efficient algorithm to estimate this metric, which adds only linear computational overhead and maintains data privacy. Empirical results on synthetic and FEMNIST datasets show that their estimation method approximates Fed-Influence with small bias.
Federated learning allows mobile clients to jointly train a global model without sending their private data to a central server. Extensive works have studied the performance guarantee of the global model, however, it is still unclear how each individual client influences the collaborative training process. In this work, we defined a new notion, called {\em Fed-Influence}, to quantify this influence over the model parameters, and proposed an effective and efficient algorithm to estimate this metric. In particular, our design satisfies several desirable properties: (1) it requires neither retraining nor retracing, adding only linear computational overhead to clients and the server; (2) it strictly maintains the tenets of federated learning, without revealing any client's local private data; and (3) it works well on both convex and non-convex loss functions, and does not require the final model to be optimal. Empirical results on a synthetic dataset and the FEMNIST dataset demonstrate that our estimation method can approximate Fed-Influence with small bias. Further, we show an application of Fed-Influence in model debugging.